crabl

this is why it's so critical (IMO) to find ways to tune the models to produce more out-of-distribution outputs. it's incredibly easy to generate "in-distribution" text and the major labs are optimizing for this because of "safety", but the only way to generate truly creative outputs is to step in and out of the fringe.

i'm betting on the "personal pan pizza team" -> won't be long until we have a 1-person $1B company

you can also chuck a json blob into LocalStorage

we've been using this Temporal polyfill and it's been awesome so far: https://github.com/js-temporal/temporal-polyfill

i'm curious to see whether this creates a large secondary market for EoL'd devices!

We diff construction schedules! These tend to be massive Gantt charts (400-700 pages is common).

We have a shared iCloud photo album which works pretty well, and a group chat in iMessage

It's interesting to me that those fonts seem to include ink traps: curious if this has anything to do with the display tech that's used in the cockpit

the Zebra disposable fountain pens are EXCELLENT for how cheap they are, it's honestly a shame they're not refillable

Bozeman tech scene is popping off recently!

Interesting! My first thought looking at the scribble chart was "isn't this Monte Carlo simulation?" but reading further it seems more aligned with the "third way" that William Briggs describes in his book Uncertainty[1]. He argues we should focus on direct probability statements about observables over getting lost in parameter estimation or hypothesis testing.

^[1]: https://link.springer.com/book/10.1007/978-3-319-39756-6

Ian Penman wrote a fantastic biography of Satie, published earlier this year. Worth a read! He was a profoundly strange and fascinating person: https://mitpress.mit.edu/9781635902532/erik-satie-three-piec...

Sole prop registrations (for individuals/non-corporations) are not _too_ onerous anymore and generally have a faster time to approval, provided you submit the right docs. Totally agree it's a money grab though. Twilio et al. have so much power to keep bad actors at bay but keep pushing responsibility to the consumers of their API. I feel like setting up push notifications is easier though, which is saying something.

Everything old is new again: I came across a demo for Telescript [1] the other day that would not look out of place in a pitch deck today, save the references to AT&T. https://www.youtube.com/watch?v=wtrs3jtY96k

[1] http://www.datarover.com/Telescript/Documentation/TRM/chapte...

As the marginal cost of writing code decreases, the opportunity cost associated with writing the "right" code increases dramatically

As a potential solution, do you think formal/semi-formal software development education (undergrad programs, colleges/polytechnics, dev bootcamps, etc) should lean super heavily into AI? To the extent that it's not just "use ChatGPT to help you complete this assignment" but rather "complete this assignment using *only* ChatGPT: you're not allowed to write any of the code by yourself".

We've been using Markdown tables to return data to the LLM with some success

What I've noticed from my extensive use over the past couple weeks has been Claude Code really sucks at thinking things through enough to understand the second and third order consequences of the code that it's writing. That said, it's easy enough to work around its deficiencies by using a model with extended thinking (Grok, GPT4.5, Sonnet 3.7 in thinking mode) to write prompts for it and use Claude Code as basically a dumb code-spewing minion. My workflow has been: give Grok enough context on the problem with specific code examples, ask it to develop an implementation plan that a junior developer can follow, and paste the result into Claude Code, asking it to diligently follow the implementation plan and nothing else.

Libraries like iText would be SO good with LLM/vision model integration and vice-versa. Huge opportunity to use these tools to generate more training data based from siloed PDFs.

ChordChord (https://chordchord.com) is close to this idea: it has an "AI suggestions" panel that recommends chords based on a progression

I feel like Kowloon is a decent metaphor for software design at most typical large SaaS companies: small changes accreted over time that lead to an impenetrable, wandering structure that only the residents (developers) truly understand.

A great book on this topic is Sweating Bullets by Robert Gaskins, the inventor of PowerPoint. It's amazing how little the core of that program has changed since it was first introduced.

TCR also recently introduced a one-time $15 A2P Campaign use case registration fee for vetting the messaging campaign details.

I worked with a similar guy: he would not do anything without first having a conversation about it. According to him, the act of writing notes based on a conversation helped concepts stick in his mind more durably than simply reading what others had written. While I'm sure there's some truth to that, it's absolutely possible to synthesize written thoughts in one's own words without directly interrogating the author. I suspect that the skill of "critical reading" is the missing link here.

https://crabl.net/

I've been working on turning my personal website into something more "app-like" (in the same vein as Brian Lovin: https://brianlovin.com/writing/how-my-website-works) and earlier this year I settled on a design that I actually don't hate for a change ;)

I have seen it in practice, and I've done it in the past: the thing to keep in mind with this approach is that it's generally most effective when the contents of the document will have a critical impact on the reader's work, and when each signatory contributed something to its creation. In the wrong context, asking for a signature can come off as distasteful or overly bureaucratic, so it's best to avoid leaning on this as a common practice.

Collecting the signatures of readers at the bottom of important documents is pretty effective ;)

In my experience (as someone who enjoys both reading and writing) it's generally a question of demonstrating the concrete value of reading and producing written artifacts. I tend to start with teaching people how to write because it provides the most tangible outcomes: an outline, a plan, a draft, or a memo. When you're able to coach someone into expressing a raw idea of theirs as words on a page, the value of writing becomes clear very quickly.

From there, it's a matter of them warming up to the notion that internalizing the writing of others works in the opposite direction. Once they understand that reading and writing is the process of serializing and deserializing mental models (and that the result of this process is lossy at best, and actively misleading at worst), that's when the switch flips from "casual" to "critical" reader, and "amateur" to "practiced" writer, in my opinion.

I don't think it's possible, although it is available on DVD/Blu-Ray, for what that's worth.

General Magic (2018) absolutely captures the essence of running & working at a startup doing innovative work on products that were well ahead of their time. Marc Porat is an incredible speaker, and he sheds a ton of light on what struggles they went through (https://www.youtube.com/watch?v=6RaTIF6st4c), and the thought process behind bringing Magic Cap to market. The soundtrack alone (done by Benji Merrison) makes the film worth watching, and it's my go-to album for when I need to get focused. Highly recommend. https://www.imdb.com/title/tt6849786/

It's interesting: we've been running our own take on the Shape Up model for a bit, and when we started out, we did not run a daily standup. After a few weeks, however, we decided to bring the group together every day: it gets lonely when working remotely, and having a daily ritual to center ourselves around is actually quite nice.